Well said, Ray.Stacking allows for (apparent) breaking the laws of physics in one dimension, while Stitching allows breaking them in the other two. And I do enjoy breaking the law.
Addicted to Stack&Stitch
Moderators: rjlittlefield, ChrisR, Chris S., Pau
- rjlittlefield
- Site Admin
- Posts: 23586
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Heh, well of course I have to quibble.Lou Jost wrote:Well said, Ray.ray_parkhurst wrote: Stacking allows for (apparent) breaking the laws of physics in one dimension, while Stitching allows breaking them in the other two. And I do enjoy breaking the law.
I grant that stacking allows for working around the laws of physics.
But stitching only allows working around the limits of engineering. There is nothing in physics that precludes a diffraction-limited NA 0.9 lens with a subject FOV of 5 inches. That lens is just really, Really, hard to build.
--Rik
-
- Posts: 1623
- Joined: Sat Apr 14, 2018 10:26 am
-
- Posts: 1959
- Joined: Sat Oct 07, 2006 10:16 am
- Location: Bigfork, Montana
- Contact:
Thanks for the explanations, the links and excellent images for comparison. I can clearly see how grading can be very controversial, especially with those that haven't done their homework. Identifying die stages and states seems to be an art in itself. I've been reading and viewing the images on varietyvista trying to understand some of this. <g>ray_parkhurst wrote:Thanks! You can tell the coin is VEDS because there is no apparent die wear. The heaviest regions of die wear on Lincoln Cents are between the rim and lettering; field at back of neck; and field in front of throat. Wear also occurs quickly at the edges of the devices where they meet the field. In the crop below, you can see no die wear at the edge where the field meets the back of head and neck:Smokedaddy wrote:... beautiful image Ray. How do YOU identify VEDS on this coin?
Interesting hobby (at this level).
-JW:
- enricosavazzi
- Posts: 1474
- Joined: Sat Nov 21, 2009 2:41 pm
- Location: Västerås, Sweden
- Contact:
Just another quibble, but the number of (unbroken) dimensions remains the same when using stacking. We get one more dimension (z) but give up time (t), which is just another dimension. We cannot stack an object that is moving reasonably fast.ray_parkhurst wrote:...
Stacking allows for (apparent) breaking the laws of physics in one dimension
...
--ES
I've been thinking about (never had actual need to do it) a concept of multiple translucent mirrors behind a single objective with different extensions between cameras and objective.enricosavazzi wrote: Just another quibble, but the number of (unbroken) dimensions remains the same when using stacking. We get one more dimension (z) but give up time (t), which is just another dimension. We cannot stack an object that is moving reasonably fast.
With telan lenses one could theoreticly capture infinitely deep stacks. Drawbacks are high cost, difference in magnification between frames, and substantial light loss (25% transmittance with 3 cameras), but this could allow to capture whole stacks simultainously and possibly produce focus stacked videos.
If one used sensor side telecentric lens, difference in magnification can be removed.
Last edited by JohnyM on Wed Sep 05, 2018 4:08 am, edited 1 time in total.
-
- Posts: 3425
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Isn't that just another engineering problem?enricosavazzi wrote:Just another quibble, but the number of (unbroken) dimensions remains the same when using stacking. We get one more dimension (z) but give up time (t), which is just another dimension. We cannot stack an object that is moving reasonably fast.ray_parkhurst wrote:...
Stacking allows for (apparent) breaking the laws of physics in one dimension
...
I propose another dimension, the $
As we've seen on the silicon semiconductor front for many years now, when enough $ are behind something, the conventional laws of physics are a barrier that becomes an obstacle rather than a limit.
Now the famous Nyquist Limit is under attack by communications systems which purposely exceed the limit to increase the data throughput, it's called Faster Than Nyquist signaling, or FTN. Why? Because bandwidth is $, and data is $, the more data pushed thru a given allocated bandwidth, the more $!! 5G is around the corner and providing the enticement (read $) to do this, although not the only enticement.
https://www.sciencedirect.com/science/a ... 0417302415
https://ieeexplore.ieee.org/document/7845669/
https://www.comsoc.org/ctn/running-fast ... -have-come
Best,
As we've seen on the silicon semiconductor front for many years now, when enough $ are behind something, the conventional laws of physics are a barrier that becomes an obstacle rather than a limit.
Now the famous Nyquist Limit is under attack by communications systems which purposely exceed the limit to increase the data throughput, it's called Faster Than Nyquist signaling, or FTN. Why? Because bandwidth is $, and data is $, the more data pushed thru a given allocated bandwidth, the more $!! 5G is around the corner and providing the enticement (read $) to do this, although not the only enticement.
https://www.sciencedirect.com/science/a ... 0417302415
https://ieeexplore.ieee.org/document/7845669/
https://www.comsoc.org/ctn/running-fast ... -have-come
Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike
~Mike
- enricosavazzi
- Posts: 1474
- Joined: Sat Nov 21, 2009 2:41 pm
- Location: Västerås, Sweden
- Contact:
We are already there, at least with the theory. Plenoptic cameras can already today take a picture of a subject that can be refocused at multiple focus points in post-processing. Using these separate images for stacking should be trivial. The practical part of the problem is that we need a sensor with a resolution of multiple gigapixels to generate images of acceptable resolution, and a large-scale parallel computer architecture to post-process the data in a reasonable time, but that is largely a logistic problem.mawyatt wrote:...
As we've seen on the silicon semiconductor front for many years now, when enough $ are behind something, the conventional laws of physics are a barrier that becomes an obstacle rather than a limit.
...
--ES
- rjlittlefield
- Site Admin
- Posts: 23586
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Sort of, but with an important limitation. It's been some years since I read the seminal dissertation, but if I recall correctly, the refocused images necessarily have much reduced resolution, compared to the shooting aperture in its plane of best focus. So, plenoptic is a great technique for say, street photography, where you might shoot at f/2 and then you're happy reconstructing at the resolution of f/16. But in the closeup/macro regime, diffraction requires us to shoot at wide aperture just to get the resolution we want, even in the plane of best focus. In that situation, the plenoptic approach doesn't help because there's nowhere for it to go.enricosavazzi wrote:Plenoptic cameras can already today take a picture of a subject that can be refocused at multiple focus points in post-processing. Using these separate images for stacking should be trivial.
--Rik
-
- Posts: 3425
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I've now created quite a few 2x3 panoramas of Lincoln Cents, with good results. I'm operating the 95PN at f3.3, which is its sharpest aperture at 2:1 magnification, for effective aperture of f9.9. This is moving somewhat into diffraction territory on my 18MP HRT2i, but I'm not seeing much visible blurring, and what is there may be helping to offset some of the demosaicing aberrations such that 100% pixel views look good/sharp/natural to me. The ~square composited panorama is ideal for coins and the overall process is fairly efficient, with a 92MP (9600x9600) final image size (cropped from ~10100x10100).
So now my question...could I do this in a single (stacked) image? What camera would I use?
I have also created 3x4 panoramas, resulting in ~150MP (12500x12500) cropped images. Can this be done in a single (stacked) image?
I'd assume the 95PN can cover this large of a format. The A version covers 64mm image circle at f3.3, so optics are covered up to this size of sensor. Since coins are round, the corner requirements are much reduced, so the lens could probably be pushed to an even larger format.
Is there a better lens for this? The 95PN is not telecentric, but I'm getting good enough performance from it. Coins are fairly flat, and while I was concerned about specular highlights shifting the look of the coin too much, with improved lighting (sectioned ringlighting) it doesn't seem to be a problem. But is there something better at 2:1?
I used the 3XMM for the 150MP panorama, and it worked well, but the final image resolution is no better (and maybe slightly worse) than the 95PN at 2x. I suppose if I could find a 5x Mag.x I'd be all set, but they don't seem to be available. Perhaps the 5x Mitty? Those seem to be the only choices from Robert's test, since the 5400 does not have enough WD for this work.
I plan to do quite a few 3x4 panoramas with the equipment as-is, I'm just trying to figure out other options for later. My camera will eventually die, though if the shutter goes I can replace it. These panoramas are pretty hard on cameras with shutters. I'm getting into the realm of you deep-stack folks.
So now my question...could I do this in a single (stacked) image? What camera would I use?
I have also created 3x4 panoramas, resulting in ~150MP (12500x12500) cropped images. Can this be done in a single (stacked) image?
I'd assume the 95PN can cover this large of a format. The A version covers 64mm image circle at f3.3, so optics are covered up to this size of sensor. Since coins are round, the corner requirements are much reduced, so the lens could probably be pushed to an even larger format.
Is there a better lens for this? The 95PN is not telecentric, but I'm getting good enough performance from it. Coins are fairly flat, and while I was concerned about specular highlights shifting the look of the coin too much, with improved lighting (sectioned ringlighting) it doesn't seem to be a problem. But is there something better at 2:1?
I used the 3XMM for the 150MP panorama, and it worked well, but the final image resolution is no better (and maybe slightly worse) than the 95PN at 2x. I suppose if I could find a 5x Mag.x I'd be all set, but they don't seem to be available. Perhaps the 5x Mitty? Those seem to be the only choices from Robert's test, since the 5400 does not have enough WD for this work.
I plan to do quite a few 3x4 panoramas with the equipment as-is, I'm just trying to figure out other options for later. My camera will eventually die, though if the shutter goes I can replace it. These panoramas are pretty hard on cameras with shutters. I'm getting into the realm of you deep-stack folks.
-
- Posts: 3425
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
The IQ4 is 14204 x 10652. So it can do a little larger than the 2x3 panorama (10652 vs 9600) but can't beat the 3x4 (10652 vs 12500). Looks like a 4:3 sensor would need to be 16668 x 12500 (208MP) to get there. The IQ4 is also 67mm diagonal, which should be fine for the 95PN given a round subject.JohnyM wrote:XF IQ4?ray_parkhurst wrote:
I have also created 3x4 panoramas, resulting in ~150MP (12500x12500) cropped images. Can this be done in a single (stacked) image?
- rjlittlefield
- Site Admin
- Posts: 23586
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
The latest sensor-shifting Hasselblad will do 400 MP, using a 100 MP sensor.
From https://www.hasselblad.com/h6d-multishot/
From https://www.hasselblad.com/h6d-multishot/
--Rik6-SHOT MULTI-SHOT MODE
In 400MP Multi-shot mode, 6 captures are made. The first 4 captures involve moving the sensor by one pixel at a time to achieve real colour data (GRGB- see 4 shot diagram below). This cycle returns the sensor to its starting point. An additional two exposures are made by moving the sensor by ½ a pixel horizontally and then ½ a pixel vertically (see 6 shot diagram).
These 6 captures are then combined to give the equivalent of a single 400MP capture, delivering a 16-bit Tiff file size of 2.4GB per frame (23200 x 17400 pixels), giving the ultimate in image resolution.